Neuron Lab

See how an artificial brain makes decisions. Start with one neuron and turn its knobs, then build a small network and watch it learn. How it works →

The neuron

b Σ Output line thickness = strength · blue = positive · orange = negative

What the neuron just did

    Example

    Inputs

    Weights · how much each input matters

    Bias · how eager it is to fire

    Activation function

    −55 10−1

    Horizontal: the sum z. Vertical: the neuron's output.

    Words you'll see

    Neuron
    A small calculator. It takes numbers in, mixes them, and passes one number out.
    Weight
    A dial for one input. A big positive weight says "this matters a lot", a negative one says "this argues against".
    Bias
    A head start added to the sum. A high bias makes the neuron say yes easily; a low one makes it hard to convince.
    Activation function
    A rule that reshapes the sum into the neuron's output, for example squashing it between 0 and 1.
    Layer
    A column of neurons that work side by side. Layers between the input and the output are called hidden.
    Training
    Showing the network examples, measuring its mistakes (the loss), and nudging every weight a little to make fewer of them.